Collection and Presentation of Data

Data Types & Sources - The Raw Material

• Data: Raw facts & figures.
• Types of Data:
  • Qualitative (Categorical): Describes qualities.
    • Nominal: Unordered categories (blood groups, gender).
    • Ordinal: Ordered categories (Likert scale, disease severity: mild/moderate/severe).
  • Quantitative (Numerical): Measurable quantities.
    • Discrete: Countable, whole numbers (number of children, admissions).
    • Continuous: Any value in a range (height, weight, BP).
• Sources of Data:
  • Primary Data: First-hand collection by investigator (surveys, interviews, experiments).
  • Secondary Data: Already collected (hospital records, census, NFHS).

    ⭐ Hospital records are a vital secondary data source for retrospective studies and health trends.

Sampling Methods - Picking Your Players

Goal: Select a representative subset from a population.

• Probability Sampling (Unbiased):
  • Simple Random (SRS): Equal chance for all (e.g., lottery).
  • Systematic: Every k-th unit selected ($k = N/n$).
  • Stratified: Population divided into homogenous strata; SRS from each. ↑Precision.
  • Cluster: Random clusters selected; all units in selected clusters sampled. Often ↑Efficiency, ↓Cost.
  • Multistage: Sampling in multiple phases (e.g., national survey).
• Non-Probability Sampling (Bias prone):
  • Convenience: Readily available participants.
  • Purposive (Judgmental): Investigator's choice based on specific criteria/judgment.
  • Quota: Predefined quotas for subgroups to mirror population proportions.
  • Snowball: Participants recruit future subjects. Useful for hidden/rare populations.

⭐ Cluster sampling: The unit of sampling is a group (e.g., village, school) rather than an individual. It is particularly useful for large, geographically dispersed populations where SRS is impractical.

Tabular Presentation - Order in the Court

• Systematic data arrangement in rows & columns; aids comparison, analysis.
• Key Parts (📌 Mnemonic: Tall Captains Shout Boldly From Ships):
  • Title: Clear, concise (What, Where, When, How classified).
  • Captions (Column heads) & Stubs (Row heads).
  • Body: Numerical data.
  • Footnote (Explanations) & Source Note (Origin).
• Types:
  • Simple (One-way): 1 characteristic.
  • Complex: Two-way (cross-tab), Three-way, Manifold.
• Principles: Logical, clear, units stated, totals, avoid clutter.

  ⭐ Cross-tabulation (two-way table) is pivotal for exploring associations between two categorical variables. A well-structured table with labeled parts like title, caption, stubs, body, footnote, and source note.)

Graphical Presentation - A Picture's Worth

• Visuals for rapid data interpretation: patterns, trends, relationships.
• Histogram:
  • Continuous quantitative data (e.g., age). Bars adjacent (touch).
  • Area of bar proportional to frequency. X-axis: class intervals; Y-axis: frequency.
• Bar Chart:
  • Qualitative or discrete data (e.g., gender). Bars separated.
  • Length of bar proportional to frequency.
  • Types: Simple, Multiple, Component/Proportional.
• Pie Chart (Sector Diagram):
  • Proportions of whole (categorical data, e.g., causes of death).
  • Total angle 360°. Sector angle = (Component value / Total value) × 360°.
• Line Chart/Graph:
  • Trends over time (time-series data, e.g., disease cases/year).
• Frequency Polygon:
  • Joins histogram tops' midpoints. Smooths data; compares distributions.
• Scatter Diagram (Correlation Plot):
  • Relationship & strength between two quantitative variables. Pattern shows correlation.
• Box & Whisker Plot:
  • Min, Q1, Median, Q3, Max. Shows data spread, central tendency, outliers.
  • Compares group distributions.

⭐ Ogives (cumulative frequency curves) graphically determine median from grouped data.

High‑Yield Points - ⚡ Biggest Takeaways

• Primary data: First-hand collection. Secondary data: Pre-existing information.
• Sampling: Simple random (equal chance), stratified (subgroup representation).
• Qualitative data: Nominal (e.g., blood type), Ordinal (e.g., pain scale).
• Quantitative data: Discrete (e.g., hospital admissions), Continuous (e.g., height).
• Presentation: Histograms for continuous data; Bar charts for categorical/discrete data.
• Frequency polygon joins histogram midpoints; Ogive (cumulative frequency) for percentiles.
• Pie charts for proportions; Scatter plots for relationships between two variables.